1 Current Account Reversals: Always a Problem?
Total Page：16
File Type：pdf, Size：1020Kb
Current Account Reversals: Always a Problem?1 Muge Adalet and Barry Eichengreen Victoria University, Wellington; and University of California, Berkeley September 2005 1. Introduction Sharp reductions in current account deficits can be disruptive. Milessi-Ferretti and Razin (2000) in their seminal study of the phenomenon, known as “current account reversals,” emphasize the dangers of large current account deficits that must be compressed when external financing dries up. Their study, written in the aftermath of the Asian crisis, presumably had countries like Thailand in mind. The authors cite other disruptive reversals, such as Uruguay’s at the beginning of the Latin American debt crisis, when financing for the current account deficit collapsed and growth fell from +5 per cent to -7 per cent.2 Looking forward there is the question of what would happen to growth in the United States if financing for the country’s 5 per cent plus current account deficit evaporated abruptly. Will the dollar fall, fanning import price inflation and forcing the Fed to raise interest rates? How would the housing and stock markets react? Sharp reductions in consumption and investment might have to be brought about by this rise in interest rates and fall in asset valuations, since the current account is the difference between investment and saving. But not all current account reversals are disruptive. In Milesi-Ferretti and Razin’s own sample, the median change in growth between the periods before and after such reversals is zero. The output response, in other words, is very heterogeneous. For every 1 Prepared for the NBER Conference on G7 Current Account Imbalances, Newport, Rhode Island, June 1- 2, 2005. We thank Gian Maria Milesi-Ferretti and Alan Taylor for assistance with data and participants in the preconference, especially Rick Mishkin, for helpful comments. 2 Between 1979-81 and 1982-84. 1 Uruguay there is a Nigeria, where growth went from -5.5 per cent in 1981-83 to +3 per cent in 1984-86 despite sharp compression of the current account. From an analytical standpoint, this is not surprising. Deficits develop for different reasons. A deficit reflecting a temporary surge in investment owing to unusually rapid productivity growth and high profitability will have different implications than a deficit reflecting a temporary surge in consumption produced by the growth of public consumption or overvaluation of the currency. Equally, current account deficits can be eliminated for number of very different reasons, which are likely to have very different output effects. Nor is it clear that current account reversals were always as disruptive as in recent years. The obvious contrast is the period before World War I, when very large deficits were allowed to develop and persist. Bayoumi (1989) considers average current account deficits over periods as long as 1870-1913 and finds that these reached high levels in countries like Australia and Canada. Taylor (1996) and Obstfeld and Taylor (2004) do the same over successive decades starting in the late 19th century and show that current account balances were larger in that period than anytime in the 20th century. To be sure, some of these deficits were compressed abruptly with interruptions to the flow of external finance, reflecting a combination of rising interest rates in the capital-exporting countries and economic and political problems in the capital importers. Instances springing to mind where current account deficits fell sharply and precipitated banking or currency crises include Denmark in 1885-6, Argentina in 1889-90, Canada in 1890-91, Australia in 1891-2, Brazil in 1896-7, Japan in 1899-1900, and Finland in 1900-1.3 Although crises 3 These are all years of or adjoining banking and currency crises as independently dated by Bordo and Eichengreen (2003). 2 can occur for reasons other than those associated with current account reversals, the connections between the two phenomena are clear. Recall Fischer’s (1988) caution that the “primary indicator [of a looming crisis] is the current account deficit.”4 At the same time, crises – currency crises in particular – were lower in frequency under the gold standard than in recent years.5 Indeed, another reading of gold standard experience is that the economic and political environment made current account reversals less of a problem. Greater wage and price flexibility in an era of unstructured labor markets facilitated the adjustment of relative prices when the current account balance had to be compressed abruptly (Bayoumi and Eichengreen 1996). With government budgets close to balance in peacetime, the twin deficits problem that gives rise to “bad current account deficits,” financing for which dries up suddenly when concerns arise about the sustainability of public debts, was less prevalent. Because large current account deficits reflected unusually high levels of investment in export-supporting infrastructure, those deficits could be smoothly reduced by increased savings out of progressively higher domestic incomes and increases in exports of goods and services (Feis 1930, Fishlow 1986).6 Since the credibility of the commitment to exchange rate stability was beyond reproach, events that might have interrupted capital inflows and forced disruptive compression of the current account elicited capital inflows that allowed that deficit to be wound down smoothly rather than precipitating a crisis. Some of these tales are consistent with fewer or smaller current account reversals, while others are consistent 4 Also cited in Edwards (2004b). 5 This is the finding of Bordo and Eichengreen (2003). 6 We can think of this as a somewhat refined version of the Lawson Doctrine as applied to the gold standard. 3 with smaller output losses (smoother adjustment to equally frequent or large current account shocks). These observations suggest a series of questions. Were current account reversals less frequent under the gold standard? Were their growth effects less disruptive? And if there are differences across epochs, what is their explanation? Bracketed by the gold standard and the post-1970 float were the 1920s and 1930s, when capital flow volatility, economic instability, and financial crises were pervasive, and the Bretton Woods quarter century, when capital flows were limited, recessions were rare, and banking crises were essentially nonexistent. Given the contingent nature of the connection between economic volatility on the one hand and current account reversals on the other, it would be illuminating to know whether reversals were larger, more frequent, and more disruptive in the interwar period – and smaller, less common, and less disruptive under Bretton Woods. In what follows we take a first cut at measuring the frequency, magnitude and effects of current account reversals in the gold standard era (1880-1914), the interwar period (1919-39), Bretton Woods (1945-70), and the post-Bretton Woods float (1972- 1997). We use regression analysis to see how far we can get in ascribing cross-period differences to observable characteristics of countries and the international economic environment. The results confirm that the gold standard era and the years since 1970 differed strikingly from one another: reversals were smaller and less frequent in the gold standard period. Controlling for, inter alia, the size of the initial current account imbalance, the movement in the real exchange rate and the state of the global economy does not make 4 this difference go away. Evidently, there was something else about the gold standard years that rendered current accounts more stable. But when reversals did take place, their effects were every bit as disruptive as after 1945. This prompts us to consider a set of case studies in an effort to shed more light on the issue. The intervening period from the 1920s through the 1960s is more difficult to characterize. The two interwar decades emerge here, as elsewhere, as years of instability: reversals were frequent and large and had major output costs. Under Bretton Woods, in contrast, reversals were few and small; in both respects this period resembles the gold standard years. These facts are presumably explicable in part by the prevalence of capital controls and tight regulation of domestic financial markets. Finally, the years since 1972 are grouped with the gold standard years in terms of ease of adjustment to reversals. The output losses from current account reversals appear to be significantly smaller not just compared to the interwar years (which is not surprising) but also compared to Bretton Woods. In the conclusion we speculate about what changes in markets and institutions might help to account for this fact. 2. The Country Sample Our empirical analysis utilizes data from Bordo and Eichengreen (2003) extended to incorporate additional variables and countries.7 The principal sources are compendia and monographs containing national historical statistics for the period prior to 1913, publications of the League of Nations for the interwar period, and standard World Bank 7 For a more extensive discussion of data sources see that publication. 5 and International Monetary Fund sources after World War II. The resulting data set has been checked and adjusted for compatibility.8 A problem for any study that undertakes historical comparisons over long periods is the country sample. Reasonably complete macroeconomic statistics including not only GDP and trade but also financial variables are available back to the late 19th century only for a subset of Western European countries, overseas regions of recent European settlement (the United States, Canada, Australia, and New Zealand), and a few of the larger Latin American countries (Argentina and Brazil). The question is whether to follow this same group of countries over time (as in for example Taylor 1996 and Obstfeld and Taylor 2004) or to add additional countries as more data become available (as in inter alia Bayoumi 1989). Both approaches have drawbacks. Following the same 10 to 15 European countries and offshoots over the entire 120 years maximizes the comparability of the country sample at the cost of representativeness.